AI Camera for Construction Sites

I’m building an end-to-end AI-driven camera system that can live on a construction site and watch every stage of the job in detail. From the first shovel in the ground to final hand-over, the camera needs to identify what is happening and who or what is involved, then log that information automatically.

Core goals
• Continuous, detailed visual analysis of the site.
• Automatic recognition and tracking of heavy machinery, hand tools and materials—essentially everything that moves or gets used on-site.
• Clean, queryable records that can later be sliced for safety audits, asset usage reports or progress reviews.

What I already have
A clear vision of the workflow and access to a live test site. Hardware selection is still flexible, so I’m open to your advice on the best camera modules, edge devices or cloud services. The important thing is accuracy, low latency and reliability in dusty, low-connectivity conditions.

What I need from you
Experience designing and coding computer-vision pipelines (Python, C++ or equivalent), familiarity with real-time object detection frameworks such as YOLO, Detectron 2 or TensorRT, and skill in integrating the model with on-device storage, an API or a small web dashboard. Please point me to previous projects that prove your experience with similar AI or site-monitoring challenges.

Deliverables
1. Model architecture, training scripts and weights.
2. Fully functional prototype that runs on the agreed hardware, recognises the specified equipment and uploads events.
3. Concise documentation so my team can maintain or expand the system.

I’m ready to move quickly once I find the right partner, so show me how your experience fits these needs and let’s get started.